JP2013511038A - Detection and correction of anomalous measurements and determination of ambiguity in a global navigation satellite system receiver. - Google Patents

Abstract

  The global navigation system comprises a first navigation receiver located in the rover and a second navigation receiver located in the base station. The single phase difference of the measured values of the satellite signals received at the two receivers is calculated and compared with the single phase difference derived from the observation model. Prior to the calculation to determine the rover output position and to determine the integer value ambiguity, an abnormal measurement is detected and removed. The detection criterion is based on the calculated and derived residual between single phase differences. In order to determine the integer value ambiguity, calculations performed based on the Cholesky information Kalman filter and the Householder transformation are preferably applied. The change in the satellite's conformation state from one epoch to another is included in the calculation.

Description

  The present invention relates generally to global navigation satellite systems, and more particularly to detecting and correcting anomalous measurements in navigation receivers and estimating ambiguities.

  The Global Navigation Satellite System (GNSS) can determine the location with high accuracy. Currently deployed global navigation satellite systems are the United States Global Positioning System (GPS) and the Russian GLONASS. Other global navigation satellite systems such as the European Galileo system are under development. In GNSS, a navigation receiver receives and processes radio signals transmitted by satellites located within the navigation receiver's line of sight. This satellite signal includes a carrier signal modulated by a pseudo-random binary code. The navigation receiver measures the time delay of the received signal with respect to a local reference clock or oscillator. Code measurements allow the navigation receiver to determine the pseudorange between the navigation receiver and the satellite. This pseudo range is different from the actual range (distance) between the satellite and the navigation receiver due to various error sources and variations in the time scale between the navigation receiver and the satellite. If signals are received from a sufficient number of satellites, the measured pseudoranges can be processed to determine code placement at the navigation receiver and to adjust the time factor. This mode of operation is referred to as a stand-alone mode because the measurements are determined by a single navigation receiver. Stand-alone systems generally provide metric accuracy.

  A differential navigation (DN) system has been developed to improve the accuracy, accuracy, stability and reliability of measurements. In the DN system, the position of the user is determined based on a base station (also called a base) whose coordinates are accurately known. The base is equipped with a navigation receiver for receiving satellite signals. The target user whose position is to be determined may be stationary or moving, and is often called a rover. The rover is also equipped with a navigation receiver for receiving satellite signals. The signal measurements processed at the base are transmitted to the rover via the communication link. For example, a communication link can be provided via a cable or an optical fiber. In many cases, the communication link is a wireless link for the convenience of a mobile rover.

  The rover processes the measurements received from the base along with the measurements obtained with its own navigation receiver to improve the localization accuracy of the position of the rover. Since the error experienced by the rover-side navigation receiver and the error experienced by the base-side navigation receiver have a very strong correlation, accuracy is improved in the differential navigation mode. Since the base coordinates are known accurately, the error on the rover side can be corrected using the measured values obtained from the base. The differential global positioning system (DGPS) calculates the location based only on the pseudorange.

  By supplementing the code pseudorange measurement with the phase measurement of the satellite carrier signal, the localization accuracy of the differential navigation system can be further improved. If the carrier phase of the signal transmitted by the same satellite is measured by both the base-side navigation receiver and the rover-side navigation receiver, the processing of two sets of carrier phase measurements will result in a few percent of the carrier wavelength. The localization accuracy can be obtained. A differential navigation system that calculates a location based on a real-time carrier signal in addition to a code pseudorange is often referred to as a real-time kinematic (RTK) system. Processing the carrier phase measurements to determine the coordinates includes an ambiguity determination step, i.e., determining the number of integer cycles in the carrier signal received by the navigation receiver from the individual satellites. It is.

  Navigation receivers (especially navigation receivers on the rover side) often operate in complex environments where measurement errors are caused by various external influences. For example, external signals can interfere with satellite signals, and structures and terrain can result in multipath errors. Errors can be classified into two broad categories: normal errors and abnormal errors. The normal error is a normally distributed white noise error, which can be corrected during the calculation of the position coordinates. Anomalous errors are large-scale system errors that can prevent accurate system position calculations. In some cases, the anomalous error may be caused by a sudden rise in intrinsic noise. Abnormal errors are more likely to occur as a result of environmental conditions. For example, anomalous errors can be caused by strong reflected signals that interfere with direct satellite signals. Similarly, extreme radio interference can also result in abnormal errors.

  The partial or complete shading of the navigation receiver can cause errors due to radio wave diffraction. Measurement errors can be minimized when the shading is partial and minor. However, if the satellite is completely shaded (ie, blocked), only the multipath signal will remain. As a result, tracking within the channel is hindered, resulting in a loss in the measured phase and thereby an abnormal error. Dynamic effects on the navigation receiver (eg certain movements of the rover) can also cause abnormal errors. Impulse acceleration affects both the receiving antenna and the crystal oscillator of the local reference oscillator, which results in intermediate carrier frequency and measured phase drift.

  One inherent type of anomalous error is a phase locked loop (PLL) cycle slip, which is a cycle slip in a PLL circuit that is tracking a satellite carrier signal. After a cycle slip occurs, the PLL circuit transitions to a new stable equilibrium point and continues to track the satellite carrier signal after that transition. If a cycle slip occurs during signal tracking, an anomalous error equal to several integer semicycles is introduced into the carrier phase measurement. If a cycle slip occurs after signal lock, an anomalous error equal to several integer cycles is introduced into the carrier phase measurement.

  Coordinate calculation from the received satellite signal requires calculation of a complicated mathematical algorithm. These algorithms are intensive calculations and often use high performance processors and memory capacity. What is needed is a method and apparatus for detecting and correcting or removing anomalous measurements that are performed prior to the execution of complex algorithms.

  The global navigation system comprises a first navigation receiver located in the rover and a second navigation receiver located in the base station. A first plurality of measurements is received based on a first plurality of carrier signals received by a first navigation receiver from a plurality of global navigation satellites. A second plurality of measurements is received based on a second plurality of carrier signals received by the second navigation receiver from the plurality of global navigation satellites. Individual carrier signals in the second plurality of carrier signals correspond to carrier signals in the first plurality of carrier signals, and individual measurements in the second plurality of measurements are the first plurality of carrier signals. Corresponds to the measured value within the measured value.

  In some embodiments, abnormal measurements are detected and then removed prior to performing calculations to determine the output position of the rover. In order to detect the abnormal measurement value, a first plurality of single phase differences based on the first plurality of measurement values and the second plurality of measurement values are calculated. A state vector based on the first plurality of single phase differences is determined. A second plurality of single phase differences based on the observation model are calculated. A plurality of residuals based on the first plurality of single phase differences and the second plurality of single phase differences are calculated. It is determined whether the first plurality of single phase differences and the second plurality of single phase differences are consistent with the observation model. Anomalous measurements are detected and removed when the first plurality of single phase differences and the second plurality of single phase differences are not consistent.

  In some embodiments, an integer value ambiguity is determined. A single phase difference between the pseudorange and the carrier phase is first calculated. These single phase differences are linearized and then the observation vectors and matrices are calculated. If the position of the rover is known, an inverse operation is performed to determine an estimate of the integer value ambiguity vector. If the rover position is not known, the ambiguity is filtered and the floating ambiguity is estimated. Integer ambiguity vector candidates are determined and determined based on fixed criteria. If the fixed criterion is satisfied, an estimate of the integer value ambiguity vector is generated.

  In some embodiments, anomalous measurements are detected and removed prior to ambiguity filtering. A fast search procedure for detecting outliers is used during the calculation. The calculation to determine the integer value ambiguity involves the generation of a Cholesky information Kalman filter and the execution of householder transformations in the Cholesky information Kalman filter.

  These and other advantages of the invention will be apparent to those of ordinary skill in the art by reference to the following detailed description and the accompanying drawings.

The flowchart of the method of detecting and correcting an abnormal measurement value is shown.

6 shows a flowchart of a method for determining an integer value ambiguity.

An expression representing initial Cholesky information coordinates is shown.

An expression representing Cholesky information coordinates in a future epoch is shown.

Fig. 7 shows pseudo code for an algorithm for updating a state vector.

An expression representing a matrix used when determining an integer value ambiguity when the reference satellite changes is shown.

Fig. 5 shows pseudo code for an algorithm used when determining an integer ambiguity when the satellite sinks. Fig. 5 shows pseudo code for an algorithm used when determining an integer ambiguity when the satellite sinks.

1 shows a high level schematic diagram of a computing system.

  The navigation measurements of interest can be considered as solutions to unknown components of the system state vector. The state vector components include the rover position vector, the rover velocity vector (if necessary), the clock error in the local reference oscillator in the base and rover side navigation receivers, and the base and rover side navigation receivers. And n integer value ambiguities of the phase measurement value of the received carrier signal. The number n will be described later. For simplicity of terminology, a navigation receiver will also be referred to herein as a receiver.

  In principle, time is measured with respect to a common system clock. In practice, each satellite transmitter and each receiver has its own local reference clock. The regional reference clock in the receiver is typically a crystal oscillator, which can experience frequency and phase variations. These variations will result in clock errors. There is also a clock error corresponding to the regional reference clock in the satellite transmitter. The clock in the satellite transmitter is generally more accurate and stable than the clock in the receiver. However, clock errors in the satellite transmitter are also taken into account in order to make accurate measurements. The difference between the time determined by the local clock in the receiver or satellite and the time determined by the common system clock is called the clock offset. In global navigation satellite systems (GNSS), discrete time scales are often used. The time referred to in connection with the time signal transmitted by the satellite is called an epoch. Another time is referenced for the digital processing circuit in the receiver.

  The carrier phase of the carrier signal is the sum of an integer number of cycles and a fractional part of the cycle. Although the fractional part of the cycle can be measured directly, the integer cycle number is initially uncertain. Therefore, it is often called integer value ambiguity. The process of determining the number of integer cycles is called ambiguity determination. Hereinafter, the carrier phase means an ambiguous carrier phase initially determined by the receiver, and the phase without ambiguity means the carrier phase without ambiguity after ambiguity determination.

For a single frequency receiver, the number n of integer value ambiguities that need to be determined is equal to the number of satellites that have signals to receive and process. However, individual satellites in a global navigation satellite system transmit signals at two or more carrier frequencies. For example, a GPS satellite can transmit signals on a carrier wave in the L ₁ frequency band and a carrier wave in the L ₂ frequency band. If the receiver is a dual frequency unit, the number n is equal to twice the number of satellites with signals to be received and processed. That is, the number n is equal to the number of satellite channels to be received and processed, in which case the satellite channels are identified by the unique characteristics of the satellite and the carrier frequency. Correction for ionospheric delay is possible by receiving signals at two carrier frequencies. These corrections simplify ambiguity determination.

  The carrier signal transmitted by the satellite follows one direct propagation path to the rover receiver and then follows the second direct propagation path to the base receiver. In this application, when both received carrier signals correspond to the same carrier signal transmitted by the satellite, the carrier signal received by the rover receiver will correspond to the carrier signal received by the base receiver. The state vector is calculated from the observation vector, and the components of the observation vector include two sets of navigation parameters. A set of parameters is determined from data measured by the base receiver, and a corresponding set of parameters is determined from data measured by the rover receiver. Each set of these navigation parameters includes the individual satellite pseudoranges to the receiver and the carrier phase of the individual satellite carrier signals. The pseudorange between the satellite and the receiver is obtained by measuring the time delay of the coded modulation signal transmitted by the satellite and received by the receiver. The encoded modulated signal is tracked by a delay locked loop (DLL) circuit in the satellite tracking channel. The carrier phase of the carrier signal is tracked by a phase locked loop (PLL) in the satellite tracking channel. These navigation parameters are determined at discrete times and then observation vectors are generated as a set of navigation parameters over a time range.

  The relationship between the state vector and the observation vector is defined by a well-known navigation simultaneous equation. Given an observation vector, if the number of equations is equal to or exceeds the number of unknowns in the state vector, the simultaneous equations can be solved to determine the state vector. To solve these simultaneous equations, use conventional statistical methods such as Least Squares (LSM), Dynamic Kalman Filter, and various modifications and combinations of LSM and Kalman methods Is also possible.

  The actual implementation of these methods in digital form can vary widely. When implementing a method in a processor, a compromise is generally made between the accuracy of the results, the speed of computation, and the load on the processor. One general scheme includes the following steps. A set of navigation parameters determined by the base receiver is transmitted to the rover. A corresponding set of navigation parameters is determined by the rover receiver. For each satellite channel at a particular time, a single phase difference between the pseudorange measured at the base receiver side and the pseudorange measured at the rover receiver side is determined, and then the base reception A single phase difference is determined between the carrier phase measured on the machine side and the carrier phase measured on the rover receiver side.

  The measured value on the base receiver side and the measured value on the rover receiver side include errors having a strong correlation. For example, the clock error in the satellite transmitter is common to signals received by both the base receiver and the rover receiver. As another example, if the base and rover are sufficiently close, the error caused by signal propagation through the atmosphere is common to signals received by both the base receiver and the rover receiver. Single phase difference corrects errors with strong correlation. That is, errors having a strong correlation are substantially eliminated.

  The residual single phase difference is then determined by subtracting the calculated value obtained from the measurement. The calculated value is based on the observation model. The remaining processing allows linearization of the initial non-linear navigation system (multiple iterations may be required). In this case, a well-known mathematical method for solving the linear equation can be applied. For example, in one embodiment, single difference measurements are first processed by a Gauss-Newton algorithm and then an initial residual is generated. These initial residuals are then processed by multiple iterations of the Gauss-Newton algorithm until convergence is reached. The final solution generates the final residue. Outlier detection and removal is performed based on these final residuals. It is also possible to use algorithms other than the Gauss-Newton method. For example, the Newton method can be used. However, this Newton method requires twice as many residual derivatives. Linearization methods during iteration include the Zeidel method and the Jacobi method. The calculation also includes searching for the minimum value of the function, and alternatively, optimization methods (such as Rosenblock method, simplex method, gradient method, and quasi-Newton method) can be used.

  The components of the state vector including ambiguity are determined. However, the calculated value of ambiguity does not necessarily have to be processed as in the case of an integer. These calculated values are often processed as floating point numbers called floating (or floating) ambiguities. The ambiguity determination process rounds the set of floating points to a set of integer values. The residual true value is then calculated from this set of integer values, and these integer values are used to again solve these simultaneous equations. A correction to the rover coordinates and the clock offset within the rover is then determined. For simplicity of terminology, the calculation of the correction value for the clock offset is also referred to as clock offset calculation.

  Calculation of coordinates and clock offsets from received satellite signals requires the calculation of complex mathematical algorithms. These algorithms are intensive calculations and often use high performance processors and memory capacity. Detection and correction or removal of unusual measurements prior to the execution of complex algorithms is advantageous to reduce processor and memory utilization. Furthermore, the coordinate calculation time is reduced and the accuracy of the coordinates is increased.

  An exceptional measurement value is also referred to herein as an abnormal measurement value. Anomalous measurements affect various processes such as position calculation and ambiguity determination. For example, if an abnormal measurement value occurs during the positioning process or the ambiguity determination process, the state vector is distorted. Abnormal measured values need to be corrected or removed first. It would be advantageous to be able to correct or eliminate anomalous measurements prior to the execution of a complex algorithm due to the computation of optimal positions and ambiguity decisions across the set of equations, which consumes considerable computational resources.

  In the embodiment of the present invention, the abnormal measurement value is detected by the abnormality detector that determines the consistency of the measurement value with the observation model. As will be described in more detail below, high-speed search and enumeration increases the calculation efficiency and reduces the number of processes, resulting in an increase in calculation speed. After an abnormal value is detected, it is possible to correct or remove the abnormal value prior to further processing of the measured value.

  FIG. 1 shows a high-level flowchart of measurement processing steps according to an embodiment of the present invention.

  Measurements 101 include pseudo-ranges, carrier phase increments over epochs, and non-ambiguity phases measured at base and rover navigation receivers. In the case of a phase without ambiguity, the initial processing to calculate the ambiguity estimate was performed in advance. In step 102, the single phase difference between the measured value measured at the rover navigation receiver side and the corresponding measured value measured at the base navigation receiver side is calculated.

  In step 104, a state vector is determined from the single phase difference. Step 104 includes correction of the pseudo-range single phase difference for the tropospheric delay and the single phase difference of the phase without ambiguity. In some embodiments, a Gauss-Newton process is used to determine the state vector. Tropospheric correction depends on the rover position and is determined during the Gauss-Newton process.

  The process then proceeds to step 106 where the residual single phase difference is calculated. As described above, the residual is determined by subtracting the value of the single phase difference calculated from the observation model from the single phase difference calculated from the measured value. In some embodiments, multiple iterations of the Gauss-Newton process are used to generate the final set of residues.

  The process then moves to step 108 where the measurement consistency is checked. The consistency criteria are described in more detail below. The process then moves to decision making step 110. If a set of measurements is consistent, the process proceeds with a rover position (specified by rover coordinates) and, if necessary, a flag (measurement value) to determine which measurement value is abnormal. An output value 111 including a flag) is generated.

  Then, if the set of measurements is not consistent, the process moves to step 114 where an outlier is detected and removed. Processing then proceeds to decision step 116. If the state vector cannot be determined from the remaining measurements, the process moves to step 118 where all measurements are removed. In order to determine the state vector, the remaining measurements need to include measurements obtained from at least four satellites. If the state vector can be determined from the remaining measurements, the process moves to step 120 where the state vector is corrected. That is, after the abnormal measurement value is removed, the state vector is recalculated. Correction can be performed using a technique that is more effective than the least squares method (LSM) performed on the remaining measurements. In one embodiment, an approach based on peer-to-peer LSM modification (which is described in further detail below) is used. The process then returns to step 106 for another iteration.

In step 108, various criteria can be used to determine whether the measurements are consistent. In this example, if the single phase difference of the measured values is consistent with the observation model, the measured values obtained from the rover receiver and the corresponding measured values obtained from the base receiver will be consistent. . Examples of consistency criteria include:
• Measurements are consistent if the absolute value of the individual residual does not exceed a user-defined threshold. This method is effective if the behavior model is known a priori.
• The measurements are consistent if the absolute value of the individual weighted residual does not exceed a user-defined threshold. The weighted residue is a residue multiplied by a coefficient called a weight value. Individual residues can be multiplied by various weights. Further details are described below.
• If the weighted sum of residual squares (WSRS) does not exceed a user-defined threshold, the measurements are consistent. This method is comprehensive because it uses all measurements. Further details of WSRS are described below.
Any combination of the above criteria can be used via the logical operations “AND” or “OR”.

In step 114, outliers can be detected using various search methods. Examples of search methods include:
• Find the maximum absolute value of the residual. Corresponding measurements are considered outliers.
• Find the maximum absolute value of the weighted residue. Corresponding measurements are considered outliers.
Search for a specific measurement such that removal of a specific measurement results in a minimum WSRS value for the updated solution. This method is a method of inspecting measurement values one at a time.

After at least one abnormal measurement value is detected, a single measurement value or a group of measurement values can be selected and designated as an abnormal value, and correction or removal can be performed. Selection methods include:
• Specify only the detected measurement values as abnormal values.
• Specify all measured values for the current channel as outliers. If the code measurement value is an abnormal value, the measurement value of the carrier phase is also designated as an abnormal value. Similarly, if the measured value of the carrier phase is an abnormal value, the code measured value is also designated as an abnormal value.
• If the measured value on the private channel is an abnormal value, the measured value on the protected channel is also specified as the abnormal value. In GPS and GLONASS, some signals can be decoded by all users, while others can be decoded only by authorized users (such as military use). In the case of GPS, the protected P-signal can be decoded by using private C-signal information.
・ Designate all measured values from the satellite where the abnormal value is detected as abnormal values.
The choice of method affects the performance of the anomaly detector.

An embodiment of an anomaly detector (as shown in FIG. 1) is described below.
・ Anomaly detector based on single phase difference in pseudo range. A single phase difference in the pseudo range is calculated. Assume that the ionospheric effects are negligible or compensated by the model. A solution of the difference code can be obtained by solving an equation related to the single phase difference in the pseudo range. WSRS can be calculated from this solution. If this sum exceeds a user-defined threshold, a decision regarding an abnormal value is made, followed by correction.
An anomaly detector based on the single phase difference of the carrier phase increments throughout the epoch. In the first epoch, rover coordinates are determined using only the code measurements to determine the pseudorange. Rover coordinate increments (coordinate changes) are determined starting from the second epoch and using the carrier phase increment throughout the epoch. The current coordinate can be determined by adding the coordinate increment value to the previous coordinate. If an abnormal value (cycle slip) does not exist, the increment value of the rover coordinate is calculated with the accuracy (cm) on the carrier phase side, but the entire rover trajectory is several tens of meters with respect to the true trajectory (in stand-alone mode) Only tens of centimeters in DGPS mode). This movement arises from inaccuracies in the initial coordinates determined from code measurements. Further details are described below.
An anomaly detector based on a single phase difference without ambiguity. A single difference between phase measurements without ambiguity is calculated. A differential phase solution can be obtained by solving an equation for a single difference in phase without ambiguity. WSRS can be calculated from this solution. If this WSRS exceeds a user-defined threshold, a decision regarding an abnormal value is made, followed by a correction.

  An example algorithm related to the detection of outliers is described below. These algorithms belong to two categories: detection of outliers during position calculation and detection of outliers before ambiguity determination. Detection of outliers during position calculation results in more accurate coordinates. Detection of an abnormal value before determining ambiguity gives a better estimate of ambiguity (this estimate also affects the calculation of the position).

Variable definitions and notation. The following variables and notations are used in the following algorithm: Vectors and matrices are shown in bold font.

Is a position vector in the World Geodetic System 84 (WGS84) coordinate frame relative to the center of the earth, and is also referred to as a radial vector in this specification.

Refers to the vector projection onto the corresponding Cartesian x, y, z axis.

-F is the index of the frequency band. (E.g., f = 1 can mean _{L 1} frequency bands, f = 2 can mean _{L 2} frequency bands). Future GNSS may have more than two frequency bands.
-S is the index of the satellite.
R is the index of the receiver (r = 0 means a rover receiver and r = 1 means a base receiver).
-K is the index of the system time.

Is the visibility pseudorange (m) between the satellite S and the receiver r. The pseudo range is also called a code measurement.

Is the measured value (m) of the visibility carrier phase between the satellite S and the receiver r. here

Represents a carrier phase represented by a cycle multiplied by a carrier wavelength.
C is the speed of light (2.9792458 × 10 ⁸ m / s).

Is the distance from the phase center of the transmit antenna on satellite S to the phase center of the receive antenna on receiver r, where:
Is the radial vector

Radial vector explaining the rotation of the earth from satellite S with

The geometric distance to the point with.
−Ω _e is the angular rotation speed of the earth (c ⁻¹ Ω _e = 2.432338779 × 10 ⁻¹³ m ⁻¹ ).

Is a radial vector of the satellite S at the time when the satellite signal is received by the receiver r. Satellite signals are transmitted at various times. In addition, a satellite signal transmitted at a single time reaches various receivers at various times. for that reason,

The value of can vary from one receiver to another.

Is a phase center displacement vector (corresponding to the frequency band f) of the receiving antenna on the receiver r relative to the antenna reference point. This displacement vector depends on the direction of the satellite S. If you write more explicitly,


Represents the phase offset (independent of direction).

Represents the phase center deviation (which depends on the elevation angle).

Represents the direction cosine from the satellite to the receiver. The second term describes the directionality of antennas operating in various frequency bands.

Is the displacement vector of the phase center (relative to frequency band f) of the transmit antenna on satellite S relative to the center of gravity of the satellite; in general, this displacement depends on the direction to the receiver r.


Represents an offset value from the center of gravity of the satellite to the transmitting antenna on the satellite.

Represents the elevation angle from the antenna plane of the satellite. And

Represents the center deviation of the satellite phase. Here, when the single phase difference of the measurement value or the time increment of the measurement value (or both) is being processed,

Note that the effects of can be ignored.
Λ ^{f, s} is the wavelength of the carrier signal transmitted by the satellite S in the frequency band f. Global Positioning System (GPS) and GALILEO (GAL) use code division multiple access (CDMA), and GLONASS (GLN) uses frequency division multiple access (FDMA). Wavelength

According to the carrier frequency ω. For GPS and GAL, the wavelength is the same for all carriers in the same frequency band f. In the case of GLN, for frequency band f,
For each satellite S, there is a corresponding satellite frequency channel number l ^s (integer). The carrier frequency for satellite S is in this case

It is. The corresponding wavelength is


It is.
This equation can be transformed into the following equation.



To maintain uniform conventions for all GNSS, similarly for channels in GPS and GAL

(This parameter is equal to 0 because the increment between satellite frequency channel numbers is 0). For the GLN frequency band:

These values are therefore the same for different frequency bands. As described above, for GPS and GAL, these values are also the same for different frequency bands because the values are all equal to zero. So for any GNSS,

Is assumed to be independent of the frequency band.
This value is

Where SYS means a specific GNSS.

Are the clock offsets of the satellite clock and the receiver clock, respectively, the clock offset relative to the system time.

Are the channel delay of the code measurement and the channel delay of the phase measurement, respectively. These delays result from radio frequency (RF) and digital signal processing of satellite signals received by the receiver. To obtain a good approximation, it can be assumed that these values are independent of satellite, time of day, and measurement format.

Is the tropospheric delay. This delay is caused by the refractive index of the troposphere.

Is the ionospheric delay caused by the propagation of satellite signals through the ionosphere. More explicitly, for a sufficiently high carrier frequency F ^{f, s} ,

Here, _ne is the electron density, L is a signal trajectory, C is a constant, and ds is a singular distance along the propagation path. Note that the frequency dependence arises from the [F ^{f, s} ] ² term. It is possible to choose any reference band that includes the exponent f _ref (eg L ₁ ), in which case


Is the ionosphere frequency ratio. Here, f _ref is an index of the reference frequency band.

Is a floating point ambiguity, where

Is an integer value ambiguity,

Is the initial phase of the satellite reference oscillator minus the initial phase of the reference oscillator of the receiver.

Is a phase incursion (phase increment) caused by a change in the orientation of the antenna on the satellite S and the antenna on the receiver r. This phase penetration occurs by changing the orientation of the antenna in the plane of the antenna's dipole axis.

Caused by mutual misalignment of the vertical axis from the line of sight to the antenna dipole

Is included.

Are the code noise error (including DLL and multipath errors) and the phase noise error (including PLL and multipath errors), respectively. These noise errors are considered white Gaussian noise including zero expected (mean) values and non-zero variance. The standard deviation (STD) value of the code noise error is about 1 m, and the STD value of the phase noise error is about 1 cm.

  Observation equation. The received satellite signal is processed by a channel algorithm to produce a raw measurement (or more precisely an estimate of the raw measurement). In some embodiments, the raw measurements are corrected. The raw measurements are then subjected to further processing depending on the particular mode of operation (stand alone, DGPS, or RTK).

The embodiment of the observation model is based on the following mathematical model. A mathematical model can be applied to the raw measurements of all Global Navigation Satellite Systems (GNSS), and this mathematical model can be formulated by the following observation equations:
and
The observation equation for the single phase difference in the pseudo range is as follows:
And the observation equation for the single phase difference of the carrier phase is as follows:
In (E3) and (E4):
• For the sake of simplicity, the single phase difference operator Δ _0,1 is omitted. For example, the single phase difference in the pseudo range is:


And the single phase difference of the carrier phase is:


• Range reference is made to the antenna reference point.
・ The following parameters are corrected (cancelled).
-Calculated distance from satellite to base-Effect of troposphere-Phase penetration (incremental value) caused by change in mutual orientation of satellite antenna and receiver antenna
– Antenna phase center offset (offset between antenna reference point and antenna phase center of rover antenna and base antenna)

Denotes the single phase difference of the ionospheric error with respect to the reference frequency.

  In (E3) and (E4), finding solutions for individual variables (more precisely, estimates of individual variables) may require additional information in some cases. However, some combinations of variables can be estimated by processing the double phase difference.

The variable double phase difference means the variable difference as a function of the satellite having the signal of interest to be received and processed. A particular satellite is selected as the reference satellite S _ref . Typically, the reference satellite is selected as the satellite with the highest elevation angle. For variables corresponding to (non-reference) satellite S, the double phase difference operator is

Represented by Then, for example,

The following variables are defined:


here,

For all values of f

Means a pair of

For all values of f and for all satellites S except the reference satellite S _ref

Means a pair. A set of equations (E3) and (E4), expressed in advance as a function of a set of variables, is


In this case, it can be expressed as a function of a set of variables:


In this case, the observation equation for the single phase difference of the carrier phase is as follows:

Is 0 for GPS and GAL and less than 1 cm for GLN. Therefore, this term can be assumed to be 0 for all systems. With this assumption, the above observation equation becomes:

And,

Hereinafter, calculation of the rover position from the code measurement value (code positioning) and calculation of the rover position from the carrier phase measurement value (phase positioning) will be considered. Short baseline ionospheric delay

Is assumed to be 0. The single phase difference observation equation for code positioning (E6) is in this case shortened to:

Ambiguity for phase positioning

Is assumed to be known and corrected. Therefore, the single phase difference observation equation for phase positioning (E7) is shortened in this case as:

Receiver clock offset. The receiver clock offset needs to be corrected. One method is to attempt to remove the clock offset from the measurement by double phase difference processing. When the double phase difference operator is applied to (E8), the following equation is obtained.

here,

Note that is removed. But,

Is present in the equation, so the noise term

Are correlated. For GPS and GAL, applying the double phase difference operator to (E9) then yields:

For GPS and GAL,

But the noise term is again correlated. In the case of GLN, the analysis is further complicated. This is because various wavelengths are involved. In that case, there are two types of double phase differences.

However, in both cases the noise term is correlated. Since the calculation of the covariance matrix and the inversion of the covariance matrix for generating the weight matrix are required, the correlation of the double phase difference is a complicated calculation. Also, as described above, the double phase difference processing for GLN provides additional processing. Therefore, the direct approach with double phase difference is not sufficient to perform receiver clock offset and phase offset removal.

  In one embodiment, the technique for performing clock offset and phase offset removal is based on a Householder transformation on the corresponding normalization equation. Normalized equation refers to an observation equation in which all terms are divided by the standard deviation (STD) of noise. In this case, the STD of normalized noise is equal to 1. This method reduces computational complexity and increases computational stability compared to the dual phase difference approach. Since the equations are pre-normalized and do not require a covariance matrix and a weight matrix, the resulting computational complexity is reduced. As the householder transformation inherently has better stability, additional computational stability results. For details on householder conversion, see C.I. L. Lawson and R.W. J. et al. Hanson's “Least Squares Problem Solving” (Prentice-Hall, Inc., Englewood Cliffs, NJ, USA (1974)) [“Lawson”]; H. Golub and C.I. F. Van Loan's “Matrix Computation” (Johns Hopkins University Press, Baltimore, MD, USA (3rd edition, 1996) [“Golub”]).

Consider the pseudorange equation (E8) for one frequency. First, the matrix and vector of the observation model including normalized noise are written.






It represents the standard deviation (STD) of the single phase difference noise of the code.

STD, also called weight root

Is a column vector containing The superscript α means the a priori position used when calculating the residual. The inverse matrix of the covariance matrix is called a weighting matrix in the case of measurement value covariance and an information matrix in the case of estimated value covariance. Here, the square root of the weighting matrix is calculated. The calculation of the single phase difference square root weight is simpler than the calculation of the double phase difference weight matrix (which is non-diagonal). In order to simplify the notation, the time index _k is omitted.

next,

Use Householder's reflection (see Lawson) to extend. This process (except for the first column element)

Is set equal to 0, resulting in:

Where NS is the number of acceptable measurements for a given frequency f. NS is determined as follows. A measured value is designated as “acceptable” if the channel has an SNR above the SNR threshold and the corresponding satellite has an elevation angle above the elevation threshold. These measurements will be shown as “acceptable” by the channel algorithm as well as by the anomaly detector. If the channel has an SNR below the SNR threshold, or if the corresponding satellite has an elevation angle below the elevation threshold, the channel or satellite measurement is “by the channel algorithm as well as by the anomaly detector. Will be shown as “unacceptable”.

The clock offset corresponds to the first row of the matrix (E13). The remaining rows of this matrix determine the normalization equation associated with the pseudo-range double phase difference. Since the operation for calculating the double phase difference in the pseudo range corresponds to the operation for multiplying the observation equation by the matrix, this result follows and sets the coefficient of the clock offset to zero. The same operation is performed, but on another matrix. As a result,

A linearized equivalent observation equation for code measurements containing is obtained. These equations are independent of clock offset.

Next, a processing procedure for removing the clock offset and the phase offset with respect to the phase positioning equation (E9) will be considered. First, the matrix and vector of the observation model including normalized noise are written.







Compared to matrix (E13), matrix (E14) is

Is included. For GPS and GAL, this column is

Is proportional to Thus, for these systems, the scheme used previously to combine the corresponding variables into one and then remove the clock offset for (E8)

It is enough to apply.

In the case of GLONASS, as described above:



And the sum of projections on additional components

Represents:

As a result,

Will result in a worse matrix condition setting than if is removed directly.

Including clock offset

Is assumed to exist. Such an estimated value can be determined, for example, from a code solution. Then in (E14)

The following equation is obtained by correcting the above.

The above expression is a vector

Including the residual error determined by. Therefore,

Can be ignored if exists within a user-defined limit.

The estimated value is within the estimated value, and the coordinate estimated value is not substantially affected.

Another way to remove clock and phase offsets is to use an additional equation from the code task to the expansion matrix (E14) as follows:
The removal of the clock offset and phase offset is performed by two successive householder transformations (E17). The first transformation resets the sub-diagonal elements in the first column (sets these elements to 0). The second transformation resets the sub-diagonal elements in the second column.
As a result, it is independent of clock offset and phase offset

A linearized observation equation representing a phase without ambiguity containing is obtained. This method of removing clock and phase offsets provides a heuristic procedure for co-processing code and carrier phase measurements.

Positioning. The positioning is realized according to a normal equation. A matrix is generated according to the following formula:


The correction for the current position is then determined:

Corresponds to the normalized double phase difference obtained from (E13), (E17), and (E18). d is called an information vector. D is called an information matrix of estimated values. D is related to the covariance matrix P as P = D ⁻¹ . The combined set {D, d} is called information coordinates.

Residual generation. In solving the problem with normalized double phase difference coupling, the coordinate estimate is determined using the double phase difference,

Is not determined. A single phase difference is defined instead of a double phase difference to correct outliers. It is important to make sure that the reference satellite outliers do not affect all of the remaining measurements. For example, the following code measurements are available for one frequency:





Assume that equation (E21) is linearized with respect to the solution of the code.


Is a partial residual vector based on geometric coordinates. (E21) is then transformed into the following equation with respect to the unknown Γ ^f :

The value Γ ^f can be estimated as follows:


And the above equation can be corrected from (E22):


Is the residual vector of the desired single phase difference. In (E23), the LSM estimate of the clock offset was used. Single phase difference residuals for other anomaly detectors are defined in a similar manner.

Outlier detection criteria. An anomaly criterion is used to identify when an outlier occurs in the current set of measurements.

It has a distribution χ ² (mn) [see Lawson]. Where m is the number of equations and n is the number of unknowns. In one embodiment, the measured value is an abnormal value under the following conditions:

Where α is the reliability of the measured value. Here, the measured value means a single phase difference of the code, a single phase difference of the phase, and a single phase difference of the phase increment value.

In another embodiment,

If is above the user-specified threshold, the measured value is an abnormal value:

Or

For example, a value of MaxWRes = 3 corresponds to “3 sigma law”.

Consider the observed linear model:

Then consider the corresponding observation model using normalized noise:

If there are no outliers,

Has a distribution χ ² (mn). Where m is the number of equations and n is the number of unknowns. Next, make two hypotheses:
H ₀ : Sum of residual squares has distribution χ ² (mn) (no outlier exists).
H ₁ : Sum of residual squares has a different distribution (there is at least one outlier).
The probability of type 1 error (also called “false alarm”) is designated as α. According to the criteria based on the χ ² distribution, the measured value becomes abnormal by the following conditional expression:

Threshold of weighted sum of residual squares (distribution fractile)

It is assumed that there is an abnormal value if it exceeds.

Abnormal measurement values can be removed according to the following method.

Remove the i-th equation from the observation model. Then find the solution of the LSM task,


The measured value with the smallest is considered an outlier. The weighted sum of the residual squares is the threshold value for removing abnormal equations

If it exceeds, the processing procedure for observation is repeated.

The search processing procedure described above is very time consuming because it is necessary to find the solution of the LSM task for each observation model. This procedure is approximately ^{O ((n 2 · m +} n 3/3) m) utilizing flop. Since m is large, there is an O (n ² · m ² ) flop. A faster method for estimating and obtaining the weighted sum of residual squares is described below. This scheme uses O (n ² · m) flops. Therefore, this method is reduced to 1 / m. This value is a considerable value for GNSS. In this case, m may be a large value of 20 or more. This approach greatly reduces the processor load and provides the same effect. In order to simplify the explanation, it is assumed that the covariance matrix of the observation noise is diagonal. The removal method used for the original observation model follows. This scheme is effective for application in addition to GNSS. One skilled in the art can apply the developed method to any observation model represented by (E28).

The LSM estimate of the state vector for the entire set of observation equations is equal to:

However,




The weighted sum of the remaining squares is:

When the i th measurement is removed

Is obtained.

And to express the estimated value through the removed measured value y _i

Indicates. The updated value of the peer-to-peer information coordinates is:



Therefore:
However,

Multiplying the matrix at (E34) and substituting these values into (E35) gives an updated value of the peer-to-peer state vector:

The peer-to-peer update of the residual square weighted sum for the removed i th equation is then:
(E37) can be simplified as follows:
This equation is obtained from the following equation:


and


Using the following formula


The following formula is then calculated:
After performing all these transformations, the weighted sum of the residual squares when the i-th equation is removed is:
Therefore, minimizing the weighted sum of the residual squares is

Is equivalent to maximization of. Since the remainder is equal to

The search algorithm is different from the selection of the absolute maximum residue in obtaining the special weighted multiplier C _(i) determined in (E35).

A search processing procedure for obtaining one abnormal value in information coordinates can be explained by the following pseudo code:
The symbol & in front of the argument variable indicates that the argument variable is an input / output variable. The value of this variable can be changed in the processing procedure. One skilled in the art can convert pseudocode to computer-executable code.

An algorithm for removing information related to an abnormal value (a high-speed recalculation of an estimated value after removing an abnormal measured value and WSRS) can be expressed by the following pseudo code.
One skilled in the art can convert pseudocode to computer-executable code. Using the procedure represented by the following pseudo code for the entire problem (inspection of measurements related to consistency, detection of anomalous measurements, removal of anomalous measurements, and repetition of the procedure for the remaining measurements) Can be solved.
A person skilled in the art can convert pseudocode to computer-executable code.

The above description considers LSM in the form of normal equations. In one embodiment, LSM is used with Cholesky's information coordinate estimate. The Cholesky information matrix (CIM) can be represented by D = S ^T S. Where S is an upper triangular matrix.

(Where x represents a state vector) is called a Cholesky information vector (CIV).
The set {S, s} is called Cholesky information coordinates (CIC).

The LSM used with the Cholesky information coordinate estimate yields:

However,

Using a computational procedure with normal equations or Cholesky information coordinates, the computation of k _(i) requires about 0.5n ² flops. Note that this is useful when the fast search procedure is linear. For non-linear observation models, an iteration can be performed on each removed equation i. In the case discussed above, the non-linearity is weak and does not require iteration.

  Ambiguity decision. In the above positioning task, it is assumed that the ambiguity has been determined. However, the single phase difference of the carrier phase is an accurate but ambiguous measurement. By converting the single phase difference of the carrier phase into a measurement value without ambiguity, it is possible to obtain a position estimation value with an accuracy of the order of 1 cm. This estimate can be obtained by the maximum likelihood method. This solution can be divided into two steps. In the first step, the floating ambiguity is determined. Various methods such as Kalman filter or recursive LSM can be used for this step. In the second step, a fixed ambiguity is determined using integer calculations.

Observation equations and state vectors for ambiguity determination. Refer back to (E8) and (E9). Assume that the variables in the first row vary independently both temporally and relative to each other. Combine equations (E8) and (E9) in the same way as when dealing with double phase differences so that these values are not taken into account. The combined equation goes to the filter input. Only the coordinates, the ambiguity double phase difference and the ionospheric error double phase difference are determined in the filter. A filtered set of floating ambiguities

An integer decision is performed on. Hereinafter, the filter will be described in more detail.

Inverse problem. Assume that the reference line (distance between base and rover) is short and the rover coordinates are known. In this case, it is possible to simplify integer determination.

Assume that the estimate has an associated error q. A known distance from the phase measurement and

And then form a double phase difference from the acquired values:

Considering the accuracy peculiar to the code measurement value, the absolute value of q can be assumed to be about 10 m or less. Therefore,

The absolute value of is less than 1 / 100th of one cycle and can be ignored. Since the phase error is much smaller than one cycle, an estimate of the integer phase ambiguity double phase difference can be obtained by rounding off the values on the left. Its value

Is less than 0.5 cycles in this case. So rounding (E46) would result in the following equation:

It is possible to determine as well. Where related to GPS and GAL measurements

Note that this value does not need to be removed since is canceled at a single phase difference. But,

If an accurate estimate of is required, the above value can be corrected. For GPS and GAL,

Cannot be estimated using only phase measurements. For this purpose,

Is estimated via code measurements.

  The flowchart of ambiguity determination. FIG. 2 is a flowchart showing the steps for determining ambiguity. [It is assumed here that there is no GLONASS bias that arises due to receivers with different characteristics (eg, receivers from different manufacturers) being installed in the rover and base. The measurements 201 include pseudoranges and carrier phases measured at the base navigation receiver side and at the rover navigation receiver side. In step 202, a single phase difference between the measured value measured at the rover navigation receiver side and the corresponding measured value measured at the base navigation receiver side is calculated. The process then moves to step 204 where the single phase difference is linearized. The process then moves to step 206 where the observation vector and matrix are determined from the linearized single phase difference. Processing then moves to decision making step 208. If the estimated value of the rover position is known, the process moves to step 210 where an inversion task that generates an estimated value 211 of integer value ambiguity is performed. For example, the rover position can be estimated from a previously known position via a single phase difference of the carrier phase increment value.

  In step 208, if the estimate of the rover position is not known, the process moves to step 220 where an anomalous measurement is detected based on the pseudo-range single phase difference and the carrier phase increment single phase difference. Removed. The process then moves to step 222 where the ambiguity is filtered. Details of step 222 will be described in more detail below. The process then moves to step 224 where the floating ambiguity is estimated. More details about floating ambiguity estimation are given below. In short, the floating ambiguity vector is a subvector of the state vector. Thus, the floating ambiguity is estimated during state vector filtering.

The process then proceeds to step 226 where integer candidate ambiguity vector candidates are determined. An integer determination is made based on the MLAMBDA method (see below). Quad form integer minimization is determined as follows.



An integer ambiguity vector,

Is the floating ambiguity vector estimated in the filter, and

Is an information matrix related to the estimate of the floating ambiguity vector. Two minimum values q ₁ ≦ q ₂ , where q ₁ is the first minimum value and q ₂ is the _second minimum value after the first minimum value and the corresponding integer value ambiguity vector

Is determined. Integer value ambiguity vector

Are selected to be candidates for integer value ambiguity vectors.

  The process then moves to step 228 where the fixed criteria for the integer value ambiguity vector candidates are evaluated. In the present application, fixed processing means processing for determining an integer value ambiguity vector from a floating ambiguity vector. This fixing process may determine an inexact integer value ambiguity vector ("wrong fixing"). Fixing standards correspond to the possibility of incorrect fixing. In some embodiments, the fixed criteria is based on a lightness ratio defined from the LAMBDA (or modified LAMBDA) algorithm output. [For details of LAMBDA and LAMBDA correction values, see X. -W. Chang, X. et al. Yang, and T.A. See Zhou's “MLAMBDA: Modified MLAMBDA Integer Least Squares Estimation Method” (Journal of Geodesy, Vol. 79, Issue 9, p. 552-565, December 2005). This document is incorporated herein by reference. ]. As the lightness ratio increases, the likelihood of incorrect fixing is reduced.

The output of the LAMBDA algorithm is an integer value ambiguity vector and a quadratic value. The lightness ratio (ratio of the value of the second minimum secondary form to the value of the first minimum secondary form) C = q ₂ / q ₁ is calculated and compared to a user-defined threshold T (T is in the vector) Depending on the number of ambiguities present in the). Fixing is achieved if C ≧ T [ie a certain integer value ambiguity vector is determined with a user-defined correct minimum probability]. Then

To be an estimate of an integer ambiguity vector. If the lightness ratio is smaller than the threshold value, the filtering process is continued to obtain a higher lightness ratio.

  Processing then proceeds to decision step 230. If the fixed criterion is met, the process yields an integer value ambiguity estimate 211. If the fixed criteria are not met, the process yields an estimate 221 of floating ambiguity. In that case, the integer value ambiguity estimate 211 and the floating ambiguity estimate 221 can be used when calculating the phase without ambiguity, and the integer value ambiguity estimate 251 Gives more accurate results than ambiguity estimates. In some embodiments, a phase without ambiguity is presented in FIG.

  For static and dynamic modes, the same method can generally be used. However, the details (filter ambiguity) of step 222 are different. Various methods can be used to perform each step. Several steps will be described below.

A series of calculations for ambiguity determination. To perform an ambiguity filter, an embodiment is used in which there is a Kalman filter with estimated Cholesky information coordinates, referred to herein as a Cholesky information Kalman filter (CIKF). Only ambiguity coordinates and double phase differences are estimated. The remaining values are removed in the process of taking the double phase difference. The ambiguity coordinates and double phase difference are contained in the state vector for the ambiguity estimation task in the form:
Only ambiguities within individual frequencies that can be defined from this observation equation are used. Thus, the size of the ambiguity subvector in (E49) is under certain circumstances, i.e. when the phase measurement is being tracked, when tracking the phase measurement and lost, and Fluctuates when the phase measurement is deleted by the anomaly detector. An important consideration is the consistency in the variable chain within the state vector from one epoch to another. Formally, the reference measurement corresponding to a specific frequency band can be removed by the anomaly detector, so there are various

Is selected.

In an embodiment of the present invention, the ambiguity determination method analyzes the current state of the satellite constellation during the transition from one epoch to another. The summary to take into account includes:
・ Reference satellite changes. The state vector and corresponding Cholesky information coordinates (CICS) are recalculated.
・ The satellite (non-reference) rises (appears). A corresponding ambiguity is introduced into the state vector.
・ The satellite (non-reference) sinks (disappears). The corresponding ambiguity is removed from the state vector.
As the satellite enters the visibility of the receiver, it rises (ie appears). That is, the receiver starts receiving a signal from a satellite that has not received a signal until then. When the satellite leaves the receiver's visibility, the satellite sinks (ie disappears). That is, the receiver stops receiving signals from the satellite that has been receiving signals. A stable solution is maintained during various changes in the satellite's conformation. Information is retained under various circumstances. In the previous procedure, information is lost, and then the computation (CIC filter) requires re-initialization and consumes computational resources. For example, when the reference satellite is changed, the entire filter is reset. A large amount of information is lost. Similar results occur when satellites appear and disappear. In some embodiments, a CIC recalculation algorithm is used to retain previously acquired information.

  As shown in FIG. 3 (E50), the Cholesky information coordinates are divided into blocks. In the next epoch, if the conformation remains the same, the CIC changes as shown in FIG. 4 (E51). Here, β∈ [0; 1] is a coordinate aging coefficient, and α∈ [0; 1] is an ambiguity aging coefficient. The ambiguity aging factor is used to model the degradation of estimation accuracy during periods when no measurements are received. (The ambiguity aging coefficient also provides a mechanism to prevent floating number overflow during program execution.) At β = 0, the coordinates change independently from epoch to epoch. β = 1 and the coordinates do not change over time (this change over time occurs when the rover is static).

Correction is performed according to the double phase difference of the linearized code observation and phase observation. The linearized observation model matrix related to the double phase difference is:

A processing procedure for updating information coordinates is used according to the observation model, except for preliminary processing of blocks for code observation. The matrix contains a number of zeros in the block row corresponding to the code observation. Therefore, the processing of 0 can be reduced. The following matrix is obtained when the update algorithm for CIKF is input.
However,

Is the Cholesky information coordinate associated with the rover position obtained from a pseudo-range double phase difference according to the following calculation scheme:
QR transforms are discussed in Lawson and Golub.

The QR transform of system [A | y] results in system [R | Q ^T y].

The update filter matrix is realized by the following processing procedure.
Here, the symbol ^~ means a predicted value, and the symbol ^ means an updated value.

Generation of double phase difference for ambiguity filters. Refer back to (E8) and (E9). The expansion matrix for this set of equations is:
In this representation, the carrier phase equation in the first equation corresponds to the reference satellite and the matrix shown below is introduced:

First, as with equation (E12), consider the block row corresponding to the code equation:

Except for the first element,

Apply a householder transformation that resets all elements of:

Then

Apply Givens rotation [see Lawson] to reset.

Again, all column elements except the first column element

Apply a householder transformation from 0 to 0.

Next, the rows are rearranged to obtain the following matrix.

The following blocks are considered to estimate the ambiguity double phase difference:

In this way, arranging the observation matrix at all frequencies associated with the ambiguity filter yields the following complete matrix:
This matrix is the same as the previously calculated matrix represented by (E52). Although it is redundant, I write down this. Failure to do so would result in a lot of work to renumber the remaining equations and drawings.

Update state vector. An update algorithm for CIC based on the calculation processing procedure (CI-LSM) will be described.

CIC a priori when entering the algorithm

Is set.

Is an observation vector including a matrix of an observation model and normalized noise. All input matrices are recalculated during algorithm processing. When output in the matrices S and s,

Is updated, then

Is reset, while

Produces a weighted residue. The update algorithm is an upper triangular matrix

Bring. The pseudo code for the update algorithm is shown in FIG. 5 (E65).

Is obtained by the backward substitution procedure [see Golub].

Is obtained by the vector S. The matrix processing “A.dotCol (i, [B,] j)” is a sequence of

return it.

Is as follows

Is rewritten: A _{:, i} : = A _{:, i} + t · B _{:, j} .

Is omitted as follows:


If the number of is 1, the argument “j” is omitted and expressed in the form of “saxpyCol”. A person skilled in the art can convert pseudocode to computer-executable code.

Change of reference satellite. Number the new reference satellite based on the frequency variable f

Is selected.

The old variable is represented using a new variable (reference satellite j) that follows the following variables:

When processing the double phase difference, the phase offset is removed by the double phase difference, so the substitution of the phase offset does not affect the task. Therefore, only the ambiguity double phase difference is important. In the form of a matrix, the ambiguity vector associated with frequency f varies according to the following rules:
Where α _{p × q} represents a matrix of the corresponding dimension, all elements of this matrix are equal to α, and n is the number of ambiguities at frequency f.

New variables are represented through old variables in a similar manner.



Where

Note that for

For the entire state vector,
In order to recalculate the CIC elements S ^x , s ^x , it should be noted here that

Please note that.



Is reduced to an upper triangular matrix using QR transform,

Derived from.

Has the following structure shown in FIG. 6 (E71).

Is the matrix A ^f and

It is calculated in consideration of the structure of A product obtained by correctly multiplying an arbitrary matrix M by a predetermined matrix A ^f is the j-th column

Is different from M only in that Matrix written in columns

in the case of,



For, it is possible to reduce the number of processes by removing the well-known 0-addend:


Applies to upper triangular and square matrices.

Since it has a special and unique structure, a multiplication result having a substantially small number of flops can be obtained. Therefore, the load on the central processing unit (CPU) is substantially reduced. In the case of an upper triangular matrix, the number of flops can be reduced to approximately one half compared to a square matrix.

In this way, the matrix (E69) is converted into an upper triangular matrix. This matrix has a partially required structure. Most of the sub-diagonal elements are zero except for the row corresponding to the old reference satellite. Diagonal block

It is sufficient to convert to an upper triangle. The structure of this block is given as follows.

Note that only the lower right block is converted to an upper triangle. To perform this conversion, the QR update algorithm [see Lawson or Golub] is used. This algorithm reduces the number of flops.
Where individual blocks

Against the block

Elements of

Note that and are recalculated. That is, the block row of the extension matrix associated with [N ^f ] is recalculated. This procedure is executed continuously for the frequency bands of the individual systems. The effectiveness of this procedure is derived from the fact that in the process of changing the reference satellite at one frequency, the element corresponding to another block row does not change. This is a characteristic of QR conversion (performed by QR update). After the reference satellite changes, the CIC is recalculated.

Ambiguity processing for sinking satellites. For a disappearing satellite, the column in the observation matrix corresponding to the satellite ambiguity is zero. Therefore, information about these columns cannot be obtained. When the satellite disappears, revocation information regarding the ambiguity of the satellite must first be removed from the CIC. This situation can occur when fading or interruptions occur on the satellite link. Non-information variables are removed from the state vector and CIC to avoid variable estimation. For this purpose, the following algorithm, expressed in pseudocode form, can be used:
A person skilled in the art can convert pseudocode to computer-executable code.

Consider the algorithm shown in FIG. 7 (E76) and FIG. 8 (E77) represented in pseudocode form. The algorithm shown in FIG. 7 forms coefficients a, α associated with a matrix that is a Cholesky decomposition on a projection matrix. By this operation, the column of S corresponding to the variable to be excluded is removed. This projection matrix can be written as:
The Cholesky matrix of the projection matrix can be expressed as:
However,
Since Γ has a special structure, the left multiplication is performed with a smaller number of flops. The above calculation is executed by a multiplication procedure.

Ambiguity processing for rising satellites. This situation occurs when a new satellite rises in the constellation. To add the above variables to the ambiguity filter state vector,

Is expanded and the corresponding row and column are filled with zeros. This means that a new variable with 0 information has been added to the state vector.

  The above-described processing procedure for handling reference satellite changes, satellite appearances and satellite disappearances is performed for all satellites and at all frequencies.

  An embodiment of a computing system that performs the above-described process for detecting and correcting outliers and then determining ambiguity is shown in FIG. The computing system 902 is typically located in a rover navigation receiver. However, the computing system 902 may be a separate unit. Those skilled in the art can configure computing system 902 from various combinations of hardware, firmware, and software. Those skilled in the art will be able to compute systems from various electronic components including general purpose microprocessors, one or more digital signal processors, one or more application specific integrated circuits (ASICs), and one or more field programmable gate arrays (FPGAs). 902 can be configured.

  The computing system 902 includes a computer 906, which includes a central processing unit (CPU) 908, a memory 910, and a data storage device 912. Data storage device 912 includes at least one persistent, non-transitory, tangible computer readable medium, such as non-volatile semiconductor memory, magnetic hard drive, or compact disk read-only memory. In the embodiment of the present invention, the computer 906 is realized as an integrated device.

  The computing system 902 can further include a user input / output interface 914 that can connect the computer 906 to the user input / output device 922. Examples of the input / output device 922 include a keyboard, a mouse, and a local access terminal. Data including computer-executable code can be transferred to and from computer 906 via input / output interface 914.

  The computing system 902 can further comprise a communication network interface 916 that connects the computer 906 with the remote access network 924. Examples of remote access network 924 include local area networks and wide area networks. A user can access computer 906 via a remote access terminal (not shown). Data containing computer-executable code can be transferred to and from computer 906 via communication network interface 916.

  Computing system 902 can further comprise a modem interface 918 that connects computer 906 with modem 926. In general, a modem refers to a communication device that receives input data from a base or another data source.

  The computing system 902 can further comprise a data bus interface 920, which includes an RF front end (which processes the RF signal received by the navigation receiver antenna), an analog to digital converter, a digital channel processing system, and Another signal processing system such as a GNSS coordinate calculation system (which calculates rover coordinates) and a data processing system 928 and a computer 906 are connected.

  As is well known, a computer operates under the control of computer software that defines the overall operation of the computer and applications. The CPU 908 controls the overall operation of the computer and application by executing computer program instructions that define the overall operation and application. Computer program instructions are stored in data storage device 912 and can be loaded into memory 910 when execution of the program instructions is desired. The method steps shown in the flowcharts of FIGS. 1 and 2 are defined by computer program instructions stored in memory 910 or data storage device 912 (or a combination of memory 910 and data storage device 912). It can be controlled by a CPU 908 that executes instructions. For example, the computer program instructions may be implemented as computer executable code programmed by a person skilled in the art to execute algorithms that implement the method steps shown in the flowcharts of FIGS. Thus, by executing computer program instructions, CPU 908 executes an algorithm that implements the method steps shown in the flowcharts of FIGS.

  It should be understood that the above-described modes for carrying out the invention are illustrative and exemplary in all respects, and are not restrictive, and the scope of the invention disclosed in this specification is as follows. It should not be determined from the detailed description, but should be determined from the claims being interpreted in accordance with the full breadth permitted by patent law. It should be understood that the embodiments shown and described herein are merely illustrative of the principles of the invention and that various modifications can be made by those skilled in the art without departing from the scope and spirit of the invention. . Those skilled in the art can implement various other feature combinations without departing from the scope and spirit of the invention.

Claims (51)

A method for processing measurements in a global navigation satellite system comprising a first navigation receiver located in a rover and a second navigation receiver located in a base station, comprising:
Receiving a first plurality of measurements based on a first plurality of carrier signals received by the first navigation receiver from a plurality of global navigation satellites;
Receiving a second plurality of measurements based on a second plurality of carrier signals received by the second navigation receiver from the plurality of global navigation satellites, the second plurality of measurements Each carrier signal in the carrier signal corresponds to a carrier signal in the first plurality of carrier signals, and each individual measurement value in the second plurality of measurement values is the first plurality of measurement values. A receiving step corresponding to the measured value in
Calculating a first plurality of single phase differences based on the first plurality of measurement values and the second plurality of measurement values;
Determining a state vector based on the first plurality of single phase differences;
Calculating a second plurality of single phase differences based on the observation model;
Calculating a plurality of residuals based on the first plurality of single phase differences and the second plurality of single phase differences;
Determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Detecting an abnormal measurement value in response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model;
A method comprising the steps of: Determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Calculating an absolute value of an individual residue within the plurality of residues;
Comparing the absolute value of an individual residue within the plurality of residues to a threshold;
When the absolute values of the individual residues in the plurality of residues are equal to or less than the threshold value, the first plurality of measurement values and the second plurality of measurement values are consistent with the observation model. Determining
When the absolute value of at least one of the residues in the plurality of residues is greater than the threshold, the first plurality of measured values and the second plurality of measured values are consistent with the observation model. A step of determining that it is not,
The method of claim 1, comprising:   The method of claim 2, wherein detecting anomalous measurements includes searching for a maximum absolute value of a residual. Determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Calculating a plurality of weighted residues based on the plurality of residues;
Calculating an absolute value of each weighted residue within the plurality of weighted residues;
Comparing the absolute value of each weighted residue in the plurality of weighted residues to a threshold;
The first plurality of measured values and the second plurality of measured values are consistent with the observation model when an absolute value of an individual weighted residue in the plurality of weighted residues is less than or equal to the threshold value. Determining that
When the absolute value of at least one weighted residue in the plurality of weighted residues is greater than the threshold, the first plurality of measured values and the second plurality of measured values are consistent with the observation model. A step of determining that it is not,
The method of claim 1, comprising:   5. The method of claim 4, wherein detecting anomalous measurement includes searching for a maximum absolute value of the weighted residue. Determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Calculating a weighted sum of squares of residuals based on the plurality of residuals;
Comparing the weighted sum of the residual squares with a threshold;
Determining that the first plurality of measurement values and the second plurality of measurement values are consistent with the observation model when a weighted sum of the residual squares is equal to or less than the threshold; ,
Determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model when a weighted sum of the residual squares is greater than the threshold; ,
The method of claim 1, comprising: The step of detecting the abnormal measurement value is as follows:
For each particular measurement within the first plurality of measurements and a corresponding particular measurement within the second plurality of measurements,
Removing the specific measurement value in the first plurality of measurement values and the corresponding specific measurement value in the second plurality of measurement values;
Calculating a weighted sum of squares of updated residuals based on the remaining measured values in the first plurality of measured values and the remaining measured values in the second plurality of measured values;
Determining a minimum value of a weighted sum of the updated residual squares;
The specific measurement value in the first plurality of measurement values and the corresponding specific measurement value in the second plurality of measurement values corresponding to the minimum value of the weighted sum of squares of the updated residuals; A step of determining
The method of claim 6, comprising:   The first plurality of measured values includes a first plurality of pseudo ranges, and the second plurality of measured values includes a second plurality of pseudo ranges. Method.   The first plurality of measurement values includes a first plurality of ambiguity-free phases, and the second plurality of measurement values include a second plurality of ambiguity-free phases. The method according to claim 1. Removing the detected anomaly measurement;
Calculating the rover position based on the remaining measurements;
The method of claim 1 further comprising: An apparatus for processing measurements in a global navigation satellite system comprising a first navigation receiver located in a rover and a second navigation receiver located in a base station,
Means for receiving a first plurality of measurements based on a first plurality of carrier signals received by the first navigation receiver from a plurality of global navigation satellites;
Means for receiving a second plurality of measurements based on a second plurality of carrier signals received by the second navigation receiver from the plurality of global navigation satellites, the second plurality of measurements; Each carrier signal in the carrier signal corresponds to a carrier signal in the first plurality of carrier signals, and each individual measurement value in the second plurality of measurement values is the first plurality of measurement values. Receiving means corresponding to the measured values in
Means for calculating a first plurality of single phase differences based on the first plurality of measured values and the second plurality of measured values;
Means for determining a state vector based on the first plurality of single phase differences;
Means for calculating a second plurality of single phase differences based on the observation model;
Means for calculating a plurality of residuals based on the first plurality of single phase differences and the second plurality of single phase differences;
Means for determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Means for detecting an abnormal measurement value in response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model;
A device comprising: Means for determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Means for calculating an absolute value of each residue in the plurality of residues;
Means for comparing absolute values of individual residues within the plurality of residues to a threshold;
When the absolute values of the individual residues in the plurality of residues are equal to or less than the threshold value, the first plurality of measurement values and the second plurality of measurement values are consistent with the observation model. Means for determining
When the absolute value of at least one of the residues in the plurality of residues is greater than the threshold, the first plurality of measured values and the second plurality of measured values are consistent with the observation model. Means for determining that it is not,
The apparatus of claim 11, comprising: In response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model, means for detecting an abnormal measurement value,
Means for retrieving a residual maximum absolute value in response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model; The apparatus according to claim 12. Means for determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Means for calculating a plurality of weighted residues based on the plurality of residues;
Means for calculating an absolute value of each weighted residue within the plurality of weighted residues;
Means for comparing an absolute value of an individual weighted residue in the plurality of weighted residues to a threshold;
The first plurality of measured values and the second plurality of measured values are consistent with the observation model when an absolute value of an individual weighted residue in the plurality of weighted residues is less than or equal to the threshold value. Means for determining that
When the absolute value of at least one weighted residue in the plurality of weighted residues is greater than the threshold, the first plurality of measured values and the second plurality of measured values are consistent with the observation model. Means for determining that it is not,
The apparatus of claim 11, comprising: In response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model, means for detecting an abnormal measurement value,
Means for retrieving a maximum absolute value of the weighted residual in response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model; The device of claim 14, wherein: Means for determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Means for calculating a weighted sum of squares of residuals based on the plurality of residuals;
Means for comparing a weighted sum of the residual squares with a threshold;
Means for determining that the first plurality of measurement values and the second plurality of measurement values are consistent with the observation model when a weighted sum of the residual squares is equal to or less than the threshold; ,
Means for determining that the first plurality of measured values and the second plurality of measured values are not consistent with the observation model when a weighted sum of the residual squares is greater than the threshold; ,
The apparatus of claim 11, comprising: In response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model, means for detecting an abnormal measurement value,
In response to determining that the first plurality of measured values and the second plurality of measured values are not consistent with the observation model,
For each particular measurement within the first plurality of measurements and a corresponding particular measurement within the second plurality of measurements,
Removing the specific measurement value in the first plurality of measurement values and the corresponding specific measurement value in the second plurality of measurement values;
Calculating a weighted sum of updated residual squares based on the remaining measurement values in the first plurality of measurement values and the remaining measurement values in the second plurality of measurement values;
Determining a minimum value of a weighted sum of squares of the updated residuals;
The specific measurement value in the first plurality of measurement values and the corresponding specific measurement value in the second plurality of measurement values corresponding to the minimum value of the weighted sum of squares of the updated residuals; 17. The apparatus of claim 16, comprising means for determining.   18. The first plurality of measurement values includes a first plurality of pseudo ranges, and the second plurality of measurement values includes a second plurality of pseudo ranges. apparatus.   The first plurality of measurement values includes a first plurality of ambiguity-free phases, and the second plurality of measurement values include a second plurality of ambiguity-free phases. The apparatus according to claim 17. Means for removing the detected abnormal measurement value;
Means for calculating the rover position based on the remaining measured values;
The apparatus of claim 11, further comprising: A computer-readable medium,
Computer program instructions for processing measurements in a global navigation satellite system comprising a first navigation receiver located in a rover and a second navigation receiver located in a base station;
Receiving a first plurality of measurements based on a first plurality of carrier signals received by the first navigation receiver from a plurality of global navigation satellites;
Receiving a second plurality of measurements based on a second plurality of carrier signals received by the second navigation receiver from the plurality of global navigation satellites, the second plurality of measurements Each carrier signal in the carrier signal corresponds to a carrier signal in the first plurality of carrier signals, and each individual measurement value in the second plurality of measurement values is the first plurality of measurement values. A receiving step corresponding to the measured value in
Calculating a first plurality of single phase differences based on the first plurality of measurement values and the second plurality of measurement values;
Determining a state vector based on the first plurality of single phase differences;
Calculating a second plurality of single phase differences based on the observation model;
Calculating a plurality of residuals based on the first plurality of single phase differences and the second plurality of single phase differences;
Determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model;
Detecting an abnormal measurement value in response to determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model;
A computer readable medium storing computer program instructions defining The computer program instructions defining a step of determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model,
Calculating an absolute value of an individual residue within the plurality of residues;
Comparing the absolute value of an individual residue within the plurality of residues to a threshold;
When the absolute values of the individual residues in the plurality of residues are equal to or less than the threshold value, the first plurality of measurement values and the second plurality of measurement values are consistent with the observation model. Determining
When the absolute value of at least one of the residues in the plurality of residues is greater than the threshold, the first plurality of measured values and the second plurality of measured values are consistent with the observation model. A step of determining that it is not,
The computer-readable medium of claim 21, comprising computer program instructions that define The computer program instructions defining a step of detecting an abnormal measurement value in response to a determination that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model Is
A computer program defining a step of searching for a maximum absolute value of a residual in response to a determination that the first plurality of measured values and the second plurality of measured values are not consistent with the observation model The computer-readable medium of claim 22, comprising instructions. The computer program instructions defining a step of determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model,
Calculating a plurality of weighted residues based on the plurality of residues;
Calculating an absolute value of each weighted residue within the plurality of weighted residues;
Comparing the absolute value of each weighted residue in the plurality of weighted residues to a threshold;
The first plurality of measured values and the second plurality of measured values are consistent with the observation model when an absolute value of an individual weighted residue in the plurality of weighted residues is less than or equal to the threshold value. Determining that
When the absolute value of at least one weighted residue in the plurality of weighted residues is greater than the threshold, the first plurality of measured values and the second plurality of measured values are consistent with the observation model. A step of determining that it is not,
The computer-readable medium of claim 21, comprising computer program instructions that define The computer program instructions defining a step of detecting an abnormal measurement value in response to a determination that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model Is
Defining a step of retrieving a maximum absolute value of the weighted residual in response to determining that the first plurality of measured values and the second plurality of measured values are not consistent with the observation model; The computer-readable medium of claim 24, comprising computer program instructions. The computer program instructions defining a step of determining whether the first plurality of measurements and the second plurality of measurements are consistent with the observation model,
Calculating a weighted sum of squares of residuals based on the plurality of residuals;
Comparing the weighted sum of the residual squares with a threshold;
Determining that the first plurality of measurement values and the second plurality of measurement values are consistent with the observation model when a weighted sum of the residual squares is equal to or less than the threshold; ,
Determining that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model when a weighted sum of the residual squares is greater than the threshold; ,
The computer-readable medium of claim 21, comprising computer program instructions that define The computer program instructions defining a step of detecting an abnormal measurement value in response to a determination that the first plurality of measurement values and the second plurality of measurement values are not consistent with the observation model Is
In response to determining that the first plurality of measured values and the second plurality of measured values are not consistent with the observation model,
For each particular measurement within the first plurality of measurements and a corresponding particular measurement within the second plurality of measurements,
Removing the specific measurement value in the first plurality of measurement values and the corresponding specific measurement value in the second plurality of measurement values;
Calculating a weighted sum of squares of updated residuals based on the remaining measured values in the first plurality of measured values and the remaining measured values in the second plurality of measured values;
Determining a minimum value of a weighted sum of the updated residual squares;
The specific measurement value in the first plurality of measurement values and the corresponding specific measurement value in the second plurality of measurement values corresponding to the minimum value of the weighted sum of squares of the updated residuals; A step of determining
27. The computer readable medium of claim 26, comprising computer program instructions that define   The first plurality of measurement values includes a first plurality of pseudoranges, and the second plurality of measurement values includes a second plurality of pseudoranges. Computer readable medium.   The first plurality of measurement values includes a first plurality of ambiguity-free phases, and the second plurality of measurement values include a second plurality of ambiguity-free phases. The computer-readable medium of claim 21. The computer program instructions for processing measurements in a global navigation satellite system comprising a first navigation receiver located in a rover and a second navigation receiver located in a base station,
Removing the detected anomaly measurement;
Calculating the rover position based on the remaining measurements;
The computer-readable medium of claim 21, further comprising computer program instructions defining: A method for determining ambiguity in a global navigation satellite system comprising a first navigation receiver located in a rover and a second navigation receiver located in a base station, comprising:
A first plurality including a first plurality of pseudoranges and a first plurality of carrier phases based on a first plurality of carrier signals received by the first navigation receiver from a plurality of global navigation satellites. Receiving a measured value of;
Receiving a second plurality of measurements based on a second plurality of carrier signals received by the second navigation receiver from the plurality of global navigation satellites, the second plurality of measurements The measured value includes a second plurality of pseudo ranges and a second plurality of carrier phases, and individual carrier signals in the second plurality of carrier signals are transformed into carrier signals in the first plurality of carrier signals. And receiving each corresponding measurement value in the second plurality of measurement values corresponding to a measurement value in the first plurality of measurement values;
Calculating a plurality of single phase differences based on the first plurality of measurement values and the second plurality of measurement values;
Linearizing the plurality of single phase differences;
Determining an observation vector and an observation matrix based on the linearized single phase difference;
Filtering the ambiguity;
Estimating a floating ambiguity based on the filtered ambiguity;
Determining integer value ambiguity vector candidates;
Determining whether the candidate integer value ambiguity vector satisfies a fixed criterion;
Generating an integer value ambiguity vector estimate in response to determining that the integer value ambiguity vector candidate satisfies the fixed criterion;
A method comprising the steps of:   32. The method of claim 31, further comprising detecting and removing anomaly measurements prior to the ambiguity filtering step. Determining whether the candidate integer value ambiguity vector satisfies a fixed criterion,
Calculating a lightness ratio;
Comparing the brightness ratio to a threshold;
Determining that a fixed criterion is satisfied when the brightness ratio is greater than or equal to the threshold;
Determining that the fixed criterion is not satisfied when the brightness ratio is less than the threshold; and
32. The method of claim 31, comprising: The step of filtering the ambiguity is
Generating a Cholesky information Kalman filter;
Performing householder transformation on the Cholesky information Kalman filter;
32. The method of claim 31, comprising: The step of filtering the ambiguity is
Selecting a specific satellite as a reference satellite from the plurality of satellites;
For individual non-reference satellites in the plurality of satellites,
Calculating a double phase difference of a user-defined variable based on the variable value corresponding to the non-reference satellite and the variable value corresponding to the reference satellite;
32. The method of claim 31, comprising:     36. The method of claim 35, further comprising determining a change in the state of the plurality of satellites from a first epoch to a second epoch. Determining a change in the state of the plurality of satellites from a first epoch to a second epoch;
Determining whether a new reference satellite has been selected;
Determining whether a new satellite has risen;
Determining whether one of the plurality of satellites sank;
38. The method of claim 36, comprising: An apparatus for determining ambiguity in a global navigation satellite system comprising a first navigation receiver located in a rover and a second navigation receiver located in a base station,
A first plurality including a first plurality of pseudoranges and a first plurality of carrier phases based on a first plurality of carrier signals received by the first navigation receiver from a plurality of global navigation satellites. Means for receiving measurements of
Means for receiving a second plurality of measurements based on a second plurality of carrier signals received by the second navigation receiver from the plurality of global navigation satellites, the second plurality of measurements; The measured value includes a second plurality of pseudo ranges and a second plurality of carrier phases, and individual carrier signals in the second plurality of carrier signals are transformed into carrier signals in the first plurality of carrier signals. Receiving means corresponding to and each individual measurement value in the second plurality of measurement values corresponding to the measurement value in the first plurality of measurement values;
Means for calculating a plurality of single phase differences based on the first plurality of measured values and the second plurality of measured values;
Means for linearizing the plurality of single phase differences;
Means for determining an observation vector and an observation matrix based on the linearized single phase difference;
Means for filtering ambiguity;
Means for estimating a floating ambiguity based on the filtered ambiguity;
Means for determining candidate integer value ambiguity vectors;
Means for determining whether a candidate for the integer value ambiguity vector satisfies a fixed criterion;
Means for generating an estimate of the integer value ambiguity vector in response to determining that the candidate for the integer value ambiguity vector satisfies the fixed criterion;
A device comprising:     40. The apparatus of claim 38, further comprising means for detecting and removing anomalous measurements prior to ambiguity filtering. Means for determining whether a candidate for the integer value ambiguity vector satisfies a fixed criterion;
Means for calculating the brightness ratio;
Means for comparing the brightness ratio to a threshold;
Means for determining that a fixed criterion is satisfied when the brightness ratio is greater than or equal to the threshold;
Means for determining that the fixed criterion is not satisfied when the lightness ratio is less than the threshold;
40. The apparatus of claim 38, comprising: The means for filtering ambiguity includes:
Means for generating a Cholesky information Kalman filter;
Means for performing householder transformation on the Cholesky information Kalman filter;
40. The apparatus of claim 38, comprising: The means for filtering ambiguity includes:
Means for selecting a specific satellite as a reference satellite from the plurality of satellites;
For individual non-reference satellites in the plurality of satellites,
Means for calculating a double phase difference of a user-defined variable based on the variable value corresponding to the non-reference satellite and the variable value corresponding to the reference satellite;
40. The apparatus of claim 38, comprising:     43. The apparatus of claim 42, further comprising means for determining a change in state of the plurality of satellites from a first epoch to a second epoch. Means for determining a change in the state of the plurality of satellites from a first epoch to a second epoch;
Means for determining whether a new reference satellite has been selected;
A means of determining whether a new satellite has risen,
Means for determining whether one of the plurality of satellites sank;
44. The apparatus of claim 43, comprising: A computer-readable medium,
Computer program instructions for determining ambiguity in a global navigation satellite system comprising a first navigation receiver located in a rover and a second navigation receiver located in a base station ,
A first plurality including a first plurality of pseudoranges and a first plurality of carrier phases based on a first plurality of carrier signals received by the first navigation receiver from a plurality of global navigation satellites. Receiving a measured value of;
A second plurality including a second plurality of pseudoranges and a second plurality of carrier phases based on a second plurality of carrier signals received by the second navigation receiver from the plurality of global navigation satellites. Measuring each of the second plurality of carrier signals, wherein each carrier signal in the second plurality of carrier signals corresponds to a carrier signal in the first plurality of carrier signals, and the second plurality of carrier signals Individual measurement values in the measurement values are received corresponding to the measurement values in the first plurality of measurement values;
Calculating a plurality of single phase differences based on the first plurality of measurement values and the second plurality of measurement values;
Linearizing the plurality of single phase differences;
Determining an observation vector and an observation matrix based on the linearized single phase difference;
Filtering the ambiguity;
Estimating a floating ambiguity based on the filtered ambiguity;
Determining integer value ambiguity vector candidates;
Determining whether the candidate integer value ambiguity vector satisfies a fixed criterion;
Generating an integer value ambiguity vector estimate in response to determining that the integer value ambiguity vector candidate satisfies the fixed criterion;
A computer readable medium storing computer program instructions defining The computer program instructions are:
46. The computer readable medium of claim 45, further comprising computer program instructions defining steps for detecting and removing anomalous measurements prior to the ambiguity filtering step. The computer program instructions defining the step of determining whether the candidate integer value ambiguity vector satisfies a fixed criterion,
Calculating a lightness ratio;
Comparing the brightness ratio to a threshold;
When the lightness ratio is greater than or equal to the threshold, determining that a fixed criterion is satisfied;
Determining that the fixed criterion is not satisfied when the contrast ratio is less than the threshold; and
The computer-readable medium of claim 45, comprising computer program instructions that define The computer program instructions defining the ambiguity filtering step are:
Generating a Cholesky information Kalman filter;
Performing householder transformation on the Cholesky information Kalman filter;
The computer-readable medium of claim 45, comprising computer program instructions that define The computer program instructions defining the ambiguity filtering step are:
Selecting a specific satellite as a reference satellite from the plurality of satellites;
For individual non-reference satellites in the plurality of satellites,
Calculating a double phase difference of a user-defined variable based on the variable value corresponding to the non-reference satellite and the variable value corresponding to the reference satellite;
The computer-readable medium of claim 45, comprising computer program instructions that define The computer program instructions are:
50. The computer-readable medium of claim 49, further comprising computer program instructions defining steps for determining a change in the state of the plurality of satellites from a first epoch to a second epoch. The computer program instructions defining the step of determining a change in the state of the plurality of satellites from a first epoch to a second epoch,
Determining whether a new reference satellite has been selected;
Determining whether a new satellite has risen;
Determining whether one of the plurality of satellites sank;
The computer-readable medium of claim 50, comprising computer program instructions that define